DEPARTMENT OF THE INTERIOR

U.S. GEOLOGICAL SURVEY

Reprocessing of Marine Multichannel

Seismic Reflection Profile Line 12,

Cape Hatteras, North Caroline, to Shell Mohawk Well, Canada

By R.A. Wise 1 and H.L. Oliver 1

Open-File Report 88-51

This report is preliminary and has not been reviewed for conformity with U.S. Geological Survey editorial standards. Any use of trade names is for descriptive purposes only and does not imply endorsement by .the U.S. Geological Survey.

'U.S. Geological Survey, Box 25046, MS 960, Denver Federal Center, Denver, CO 80225

1987

CONTENTSPage

Abstract............................................................ 1Introduction........................................................ 1Operational problems................................................ 1Reprocessing........................................................ 3Conclusion.......................................................... 6Reprocessing recommendations........................................ 6Acknowledgments..................................................... 6

TABLES

Table 1. Original processing sequence............................... 92. Reprocessing sequence...................................... 103. Library compression........................................ 11

ILLUSTRATIONS

Figure 1. Location map.............................................. 22. Salt dome, Area 1, original section ...................... 43. Salt dome, Area 1, reprocessed section ................... 54. Questionable structure, Area 2, original section.......... 75. Questionable structure, Area 2, reprocessed section....... 86. Original section, Area 3.................................. 127. Reprocessed section, Area 3, showing crisper reflections.. 138. Graben, Area 4, original section.......................... 149. Graben, Area 4, reprocessed section....................... 1510. Original section, Area 2, processed to 8.0 seconds........ 1611. Reprocessed section, Area 2, processed to 10.0 seconds.... 17

Reprocessing of Marine MultichannelSeismic Reflection Profile Line 12:

Cape Hatteras, North Carolina, to Shell Mohawk Well, Canada

By R.A. Wise and H.L. Oliver

ABSTRACT

U.S. Geological Survey marine multichannel seismic reflection profile Line 12 traversed 1,437 kilometers from Cape Hatteras to the Shell Mohawk well in Canada. Because of questionable velocity information and the poor quality of the original data processing done in 1978, reprocessing was requested. Five features were selected from the reprocessed line, all of which showed that with careful editing of data and careful management of certain key parameters in the software modules, the seismic image of grabens, salt domes, and deep crustal reflectors throughout the profile were enhanced, enabling a more accurate interpretation of the data.

INTRODUCTION

U.S. Geological Survey marine seismic multichannel reflection profile Line 12 (fig. 1) was shot in 1975 as part of a three-year data acquisition project to give the USGS its first deep look at the arrangement of sedimentary rock layers in the mid to north Atlantic basins. Seismic Line 12 intersects 27 dip profiles and is the only regional strike line from Cape Hatteras to the Shell Mohawk well acquired in less than 100 meters of water. This particular data set was collected by Digicon Geophysical using their vessel, the Gulf Seal. The data collected were recorded by a DPS III recording system, using a 48-channel, 3,600 meter, nonlinear streamer. The seismic source was a tuned airgun array totaling 1,700 cubic inches. The geometric configuration of the streamer and the source interval was designed for 36-fold processing. Data was recorded to 10 and 7 seconds at a 2-millisecond sampling rate. Reprocessing of this data was done for the following reasons:

1. To eliminate multiples and data ringing that dominated the original profile causing unreliable picks of primary reflections.

2. To reveal and enhance primary reflections that were not seen in the original processed profile, or were highly contaminated by multiples, ringing, or random noise.

3. To provide accurate velocity information along the profile.4. To accurately tie intersecting profiles.

OPERATIONAL PROBLEMS

The demultiplexing of marine seismic multichannel reflection data Line 12 caused several operational problems. These problems relate entirely to the original field tapes. The major problem with these field tapes is a tape skew condition. This condition occurred during the original recording of the data. During the entire survey, only one of the two tape drives used to record the data had this skew problem. Skew occurs when magnetic tapes are not mounted properly with respect to the magnetic recording heads. This causes the recording tracks to deviate

IBAL

TIMO

RE C

ANYO

N TR

OUGH

AT

LA

NT

IC C

ON

TIN

EN

TA

L M

AR

GIN

Figu

re 1

.~ L

ocat

ion

map

.

from their proper positions and produces crossfeed and parity errors. Due to this skew problem, more than twice the normal time was needed to complete the demultiplexing process, and even with the experienced senior computer operator, approximately 1,460 of the 29,200 shots along the line were lost.

The second major problem with the field tapes was a partial deterioration of the tapes themselves. This problem occurs when the magnetic part of the tape cracks or chips. This problem results in either a total loss of the shot if the record header ID was damaged due to deterioration, or data dropout when an area the size of a needle point is corrupted in the data itself. Deterioration is a problem that occurs with age and storage conditions of field tapes. There should be real concern with older field tapes, as we are experiencing data dropout reflected in shots lost along older lines.

REPROCESSING

Once original field tapes were properly demultiplexed, reprocessing proceeded at an efficient rate. Shot records were carefully edited to eliminate traces that contained noise, due to bad hydrophones in the streamer, or corruption in the recording system. Using parameters within the software edit-module with care and accuracy eliminated bad shot records that could have affected CDP (common depth point) records after sorting.

After edit, shot records were processed into CDP sorted records. Testing was performed to determine a mute pattern, deconvolution parameters, and filter response to eliminate and attenuate noise and to attenuate multiples and ringing. It was observed that original processed data had a very shallow mute that allowed noise in the form of refractions and direct arrivals to contaminate the stacked profile. Careful observation of CDP records allowed accurate parameters in the software mute-module to be applied, eliminating these forms of noise.

Deconvolution before stack was ignored completely in the original stacked data. This can be seen immediately (fig. 2) by the strong multiples and data ring. By testing deconvolution before stack, we found that a spiking deconvolution using an operator length of 124 milliseconds and two or three filter windows, depending on total length of data, attenuated multiples and data ring. By observation of CDP records and autocorrelations, parameters within the software deconvolution-module were made (fig. 3).

As original velocity information was not available, new velocity information was generated. This information was input, and normal moveout (NMO) was applied. A final mute was designed after NMO that eliminated upward effects of noise, data stretch, and first break noise suppression. The data was then stacked 36 fold.

Post-stack deconvolution was tested and it was determined that a 2nd zero crossing deconvolution using an operator length of 400 ms and two filter windows would collapse the wavelets into a more distinct reflection. Final post-stack filters were tested and a time variant filter was designed (see processing sequence). An automatic gain (AGC) of 500 ms was applied to the data along with a post-stack water-bottom mute and the data was displayed.

TIME IN SECONDS

1. 2.

Figu

re 2. Seismic ex

ampl

e of

original stacked

profile

show

ing

mult

iple

s,

data

ri

ng,

and

nois

e ca

used

in part by

no

de

conv

olut

ion

before'

stack.

Refl

ecto

rs overlap

and

run

into ea

ch ot

her

caus

ing

interpretation to

be difficult.

Exam

ple

is fr

om Ar

ea 1

on th

e location ma

p.

Direction

is ea

st to

th

e ri

ght.

Sc

ale

is

2"=2

.5 ki

lome

ters

.

TIME

IN SE

COND

S Figu

re 3. Same ex

ampl

e as

fi

gure

2

afte

r re

proc

essi

ng.

Note ho

w th

ere

flec

tors

are

crisp

and

defined.

Multiples

and

data ri

ng have

been attenuated.

Improvements ar

e du

e to

ca

refu

l ed

itin

g and

deconvolution

befo

re st

ack.

Several areas of structure at a time of 1.6 to 2.4 seconds, observed in the original section (fig. 4), were nonexistent in the reprocessing (fig. 5). These particular areas were reworked with extensive testing of geometry, deconvolution, muting, filtering, and velocities, with results confirming the accuracy of the reprocessed section. It was concluded that the original was in some way corrupted.

Table 1 describes the original processing flow. Table 2 describes the reprocessing flow.

CONCLUSION

The reprocessing of the USGS marine seismic multichannel reflection profile Line 12 demonstrates how standard processing can result in a superior product by careful editing of the data and testing of the software modules to determine parameters, prior to final summing of the data. Multiples and data ring that dominated the original summed data were attenuated in the reprocessing. Noise that corrupted the original section was almost eliminated in the reprocessing. All of the goals for reprocessing the data were achieved; figures 6-11 demonstrate the substantial improvement achieved through the reprocessing.

In addition to the reprocessing improvement, we were able to increase our data library storage capacity by 572 slots. This was accomplished by outputting demultiplexed data in a SEG-Y format at 6250 BPI and eliminating the original field data. Table 3 describes the results obtainable from this procedure.

REPROCESSING RECOMMENDATIONS

It is suggested that questionable marine multichannel seismic reflection profiles acquired by the USGS and processed by the USGS prior to 1979 be reprocessed. Using this profile as an example, the probability of superior results are high.

Many specific areas along profile line 12 warrant special processing. Areas of structure can be cleaned, not only attenuating but eliminating multiples. With our "discovery" system, horizons and structures can be modeled in 3 dimensions. Our group has the ability to provide some very powerful and exciting modeling displays using these older data sets.

ACKNOWLEDGMENTS

The following people provided invaluable assistance in completing this part of the project:

Warren Agena* John Grow* Myung Lee* John Miller* Nick Zihlman*

*Seismic Stratigraphic Group, Denver, CO

2.1

^O^W^j^e^ 'asf^wfc^^^

3.1

;^^'.^^^:7?.i' ^*^fow.^^^^ v

jFigure 4. Example of original stacked profile from Area 2 on the location*

map. Strong reflector at 2.0 seconds showing dip to the northeast has been questioned. Direction NE to the right. Scale 2"-2.5 kilometers.

TIME IN SECONDS Ml

Figure 5.--Same example as figure 4 after reprocessing. Dip to NE is nowminimal. Overall profile is crisper and better defined due to __ editing and deconvolution, more accurate velocity information, «~-« and better designed time variant filter application "*"" (preservation of higher frequencies).

Table 1. Original processing sequence

PROCESSING PARAMETERS

CORRECTIONS

DATUM Sea Level

VE VW

AUTOMATIC STATICS WINDOW

ADDITIONAL

TO SEC

LENGTH (MS) __WINDOW (SEC) PREDICTION (MS)

DECONVOLUTION BEFORE STACK

NONE

DECONVOLUTION AFTER STACK 240 0.3-8.0 50

BAND PASS

BAND-PASS FILTERING 12-50

8-43

4-30

SAMPLE RATE: 4 MS

ONE INCH: 24 TRACES

ONE SECOND: 2.5 INCHES

PLAYBACK GAIN: 3 DB

APPL. (SEC) OVERLAP (SEC)

0.0-1.4 .140

1.4-3.0 .160

3.0-8.0

MEAN VALUE: 7,200

Table 2. Reprocessing sequence

PROCESSING SEQUENCE1 DEMULTIPLEX2 RECORDING GRIN REMOVRL3 GEOMETRY DEFINITION4 TRRCE EDITING5 RESRMPLE TO 4.0 MS6 COMPRESS TO 36 TRRCE 100 METER RECORDS7 CDP SORT 36 FOLD8 RUTOMRTIC GRIN CONTROL

GRTE LENGTH: 1000 MS9 VELOCITY RNRLYSIS10 NORMRL MOVEOUT CORRECTION11 FIRST BRERK NOISE SUPPRESSION (MUTE)12 PRE-STflCK DECONVOLUTION

TYPE: SPIKING OPERRTOR LENGTH: 31 POINTS TIME WINDOW: 0 - 2000 TYPE: SPIKING OPERRTOR LENGTH: 31 POINTS TIME WINDOW: 3000 - 5000 TYPE: SPIKING OPERRTOR LENGTH: 31 POINTS TIME WINDOW: 6000 - 10000

13 STfiCK 36 FOLD14 POST-STRCK DECONVOLUTION

TYPE: 2ND ZERO CROSS. OPERRTOR LENGTH: 100 POINTS TIME WINDOW: 0 - 3000 TYPE: 2ND ZERO CROSS, OPERRTOR LENGTH: 100 POINTS TIME WINDOW: 4000 - 10000

15 BRNDPR5S FILTERTIME: BRNDPRSS:

0 - 500 MS 8/12 - 40/45 1500 - 30130 MS 8/10 - 35/40 5000 - 10000 MS 5/8 - 28/33

16 RUTOMRTIC GRIN CONTROLGflTE LENGTH: 500 MS

17 DI5PLRYSHOT POINTS RDJUSTED TO RCTUflL flNTENNR POSITION

10

Table 3. Library compression

Cost Efficiency

756 > Field Tapes Input 800 BPI Field Tapes

187-

Demultiplex

Output 6250 BPI SEG-Y Archive Tapes To Replace Original Field Tapes

OUTPUT RATIO 1:4.1

11

gure Example of original stacked profile from Area 3. Clarity of reflectors is poor due to velocities, lack of deconvolution, and filtering. Compare the data at 2.4 to 3.4 seconds with figure 7. Direction is east to the right, and scale is 2"=2.5 kilometers.

12

TIME IN SECONDS -. .!»,

1.600

jjfostfs.ig.'y' '^'!^{g?.><a=o^*^yg^ "

1.700

1.800

1.900

2.0C2. 1C

2.200 55 *>w>K7

2.300

2.400

S^S^E^^

»'Mr/«oy/-^^Sj«504/irfKA:^«tr^-«t^*npi' ' « ***^«£v: '^t»««««iit^rnmim

«J»^^

2.:

2.

2.

2.800

iiiiiiiia^^^i*^^6au^<«sc<£;<im^«f_. . - f*-^^i^^t^;:A'»^^^r:r^^^

3.3. 1

3.

3.300

3.400

3.

3.600

3.700

3.800

3.900

4.0

4. 100

4.200

4.300

4.400

4.500

4.600

4.700 .

21SM5 «*iMiJ»*.i«'K»*.'- «r^?*i?^|>!il4«»'«iS5»*'^<*«3i?5«* 7artdE^X%̂ &i^b^v/j^w^i£«i.x. ^^C^V&^J*?sscsssaiagsaais^^.-....._ ...........___. _. _........._......._..__ --£

^^^^^^s^^^^^^^^^^^^^S

:i-».«»<

»o3 ts^^^CIS^j^pf^^;^^^^^^^%

Figure 7. Same example as figure 6 after reprocessing. More reliable interpretation of faults and structure. Reflectors stand out due to careful processing of the data.

13

TIME IN

SE

COND

S

V'»v

^^^>

>«.M

*«<t

o«^n

«KA

*MrjA

g4g

*"<*

^ww

jte^^

*-««

i^^J

^A

^S

^^^^A

^^^^^IA

'^^^M

^

rv^

**'K

»i\n

i^l»

/wm

v^W

«V:*

*--.^

- *-

- '' *

* »*

'fi

'wf^

iWIH

&St

fSi

f**W

MW

Wo

-VfM

l

«3

MM

fcW

MM

MX

AiU

M '

.

*HW

I4»M

NM

:*

^'VW

.

* ,^

tiiifc

Mli>

v'l^V

i*^r

>>Lj

JW?'

s.i

" '-«^

"j-^^

««.H

wmSi

Sspc

" l^

Wir^

» v k »_i*

^«

m»it

jv ,A

4J Fig

ure 8. Example of or

igin

al stacked

profile

from

location ma

p area 4

across a

grab

en.

Frequency

of data is

ve

ry low

corrupting th

e detail of

th

e structure.

Reflectors run

into

each ot

her

allo

wing

for

difficult

interpretation.

Direction

is northeast

to th

e right.

Scal

e 2"

=2.5

kilometers.

Grab

en

Interpretation

Same

example

aj

better defined.

Clean

becomes

more

reliable.

Figu

re 10

. Or

igin

al pr

oces

sed

stac

ked

profile.

Note

or

igin

al pr

oces

sing

wa

s to

8.0

seconds

only.

Poor choice of

parameters ca

used

th

is

section

to ha

ve the

same

ov

eral

l lo

ok.

Nois

e and

velo

citi

es

have

corrupted

this

data.

Ther

e was

no deconvolution

befo

re

stac

k ap

plie

d.

Location map

Area 2.

Di

rect

ion

is ea

st to

th

e right.

Scale

2"=2

.5 kilometers.

TIME

IN SECONDS

>6&'<^^I^S!MU<!C*MtWrWfriW^

Figu

re 11.

Full

10 se

cond

reprocessing us

ing

new

velocity an

alys

is,

deco

nvol

utio

n be

fore

stack, and

a we

ll-d

esig

ned

time

va

rian

t fi

lter

have im

prov

ed th

e ov

eral

l se

ctio

n.

A fi

ne re

flec

tor

af

9.0

seco

nds

was

reve

aled

us

ing

all

the

data

length.